Resolution

1. Overview

The Meteorological Service of Canada (MSC) Open Data ecosystem provides a diverse range of data products that are essential for high-accuracy heatwave prediction. These products vary in both spatial and temporal resolutions, allowing us to capture detailed localized phenomena (such as urban heat islands) as well as broader regional and national weather patterns. Below is a breakdown of the primary data sources, their spatial and temporal characteristics, and how they integrate into our system.

2. Data Sources and Their Resolutions

2.1 Numerical Weather Prediction (NWP) Forecasts

High Resolution Deterministic Prediction System (HRDPS):

  • Spatial Resolution: Approximately 2.5 km grid spacing

  • Temporal Resolution: Forecasts are typically updated every hour, with output available at sub-hourly intervals (e.g., every 15–30 minutes) via interpolation.

  • Usage: Provides detailed, local-scale forecasts crucial for urban environments like Toronto.

Regional Deterministic Prediction System (RDPS):

  • Spatial Resolution: Approximately 10 km grid spacing

  • Temporal Resolution: Hourly updates with potential sub-hourly products for vertical profiles and mesoscale phenomena.

  • Usage: Useful for capturing broader mesoscale features that influence heatwave formation across larger regions.

Global Deterministic Prediction System (GDPS):

  • Spatial Resolution: Typically 20–40 km grid spacing

  • Temporal Resolution: Updated hourly, providing a global context for large-scale pressure systems and atmospheric patterns.

  • Usage: Serves as boundary data for regional models and provides context for ensemble forecasting.

Ensemble Prediction Systems (e.g., GEPS, REPS, NAEFS):

  • Spatial Resolution: Generally aligned with their deterministic counterparts (e.g., 20–40 km for GEPS)

  • Temporal Resolution: Hourly updates, offering multiple forecast members to quantify uncertainty.

  • Usage: Essential for probabilistic forecasting and uncertainty quantification.

2.2 Observational Data

In Situ Observations (Ground Stations):

  • Spatial Resolution: Point measurements at individual station locations (e.g., Toronto Pearson, local weather stations)

  • Temporal Resolution: Typically available on an hourly basis, with some stations offering sub-hourly (e.g., 10-minute) updates.

  • Usage: Provides ground-truth data for model validation and calibration.

Weather Radar Imagery:

  • Spatial Resolution: High resolution, typically around 1 km (e.g., GeoTIFF 1km radar composite)

  • Temporal Resolution: Updated frequently—often every 5 to 10 minutes—to capture rapid developments in precipitation and convective activity.

  • Usage: Critical for monitoring storm development, convective initiation, and capturing localized precipitation that affects surface cooling.

Satellite Observations:

  • Spatial Resolution: Varies by product; often around 1 km for many visible and infrared imagery products

  • Temporal Resolution: Can range from 15 minutes (for high-temporal resolution geostationary satellites) to hourly updates.

  • Usage: Used for broad-scale monitoring of cloud cover, land-surface temperatures, and vegetation indices.

2.3 Climate Data

Historical Climate Records (AHCCD, CANGRD):

  • Spatial Resolution: Often available as gridded datasets with resolutions ranging from 5 km to 25 km

  • Temporal Resolution: Typically provided as daily, monthly, or seasonal averages

  • Usage: Essential for long-term trend analysis and for training models on historical heatwave events.

Global Climate Model Scenarios (CMIP5/CMIP6):

  • Spatial Resolution: Coarser, generally 50 km or more, depending on the model

  • Temporal Resolution: Monthly to annual projections

  • Usage: Useful for understanding potential future shifts in heatwave frequency and intensity over long time horizons.

2.4 Supplementary and Derived Data

Urban and Socioeconomic Data:

  • Spatial Resolution: Typically at the city block or census tract level for urban features (e.g., land use, building density, greenspace)

  • Temporal Resolution: Updated as per municipal datasets (often annually or less frequently)

  • Usage: To capture the urban heat island effect and to correlate heatwave impacts with socioeconomic factors.

Derived Indices (Calculated In-House): Using the raw data, we derive key indices such as:

  • Heat Index (HI): Combining temperature and humidity.

  • Wet-Bulb Globe Temperature (WBGT): Accounting for temperature, humidity, and radiant heat.

  • SPEI (Standardized Precipitation Evapotranspiration Index): Reflecting moisture deficits.

  • CAPE and CIN: Calculated from vertical atmospheric profiles to assess instability.

These indices are computed at the same temporal resolutions as the underlying data (hourly or sub-hourly for real-time forecasting and daily or monthly for climate trend analyses) and are spatially interpolated to match the resolution of our high-resolution forecasts.

3. Summary of Spatial and Temporal Specifications

Data Category

Spatial Resolution

Temporal Resolution

Primary Use

HRDPS (NWP)

~2.5 km grid

Hourly (15-30 min interpolated)

Local, high-resolution forecasts

RDPS (NWP)

~10 km grid

Hourly

Mesoscale forecasting

GDPS (NWP)

20–40 km grid

Hourly

Global context, boundary conditions

Ensemble Forecasts

~20–40 km grid (varies)

Hourly

Uncertainty quantification

Ground Observations

Point measurements

Hourly (sub-hourly available)

Model calibration and validation

Weather Radar Imagery

~1 km (GeoTIFF composite)

5–10 minutes

Convective activity and precipitation

Satellite Imagery

~1 km

15 minutes to 1 hour

Broad-scale monitoring (clouds, LST)

Historical Climate Data

5–25 km (gridded)

Daily, monthly, seasonal

Long-term trend analysis

Global Climate Scenarios

~50 km or coarser

Monthly to annual

Future projections and trend analysis

Urban Data

City block to census tract

Annually (or less frequent)

Urban heat island assessment

4. Implications for Model Development

  • High Temporal Resolution: The availability of sub-hourly to hourly updates is critical for capturing rapid changes in temperature and humidity that lead to heatwaves. This frequency supports dynamic, near-real-time forecasting and timely alerts.

  • High Spatial Resolution: With grid resolutions as fine as 2.5 km (HRDPS) and radar composites at 1 km, our models can capture localized phenomena like urban heat islands. This granularity is vital for accurately predicting heatwave impacts in densely built environments like Toronto.

  • Historical and Climate Data Integration: Daily to monthly climate records and global model scenarios provide the necessary context to train models on long-term trends and extreme events, ensuring that predictions are both accurate in the short term and robust over extended periods.

  • Derived Features and Indices: Calculating indices such as Heat Index, WBGT, SPEI, CAPE, and CIN using the above data sources allows our models to quantify heat stress and atmospheric instability effectively, enabling more precise forecasting.

5. Conclusion

The MSC Open Data ecosystem provides an extensive suite of high-resolution data products essential for our heatwave prediction system. With spatial resolutions ranging from point measurements to grid-based forecasts (as fine as 2.5 km) and temporal resolutions spanning from 5-minute radar updates to hourly observations and daily climate records, we have the necessary granularity to capture both local and broad-scale phenomena. These specifications enable our AI/ML models to generate highly accurate, real-time forecasts and early warnings, ultimately supporting robust, risk-informed decision-making for urban and regional resilience.

PreviousAppendix A: Data SourceNextSurface

Last updated

Was this helpful?