Charter 1: Explore forecast database with reference to latest air quality forecasts for the list of the CNN cities to discover if pollutant values are correct for given cities.

Charter:

Explore forecast database with reference to latest air quality forecasts for the list of the CNN cities to discover if pollutant values are correct for given cities.

Testers:

• Mike Walker-Rose

Date:

• 17/05/24

Duration:

• 2 hours (approx. 15:00 – 17:00

Scope:

• Testing the forecast database to establish if the forecast data is correctly extracted from CAMS and associated to the specified cities • Use the ECMWF’s forecast as an oracle to measure forecast accuracy, record differences

References:

• List of stations (https://myftp.ecmwf.int/view/public/cams/products/cams_global_forecast/surface_concentrations/CAMS_locations_V1.csv) • Latest air quality forecasts for the list of the CNN cities in CSV format (https://myftp.ecmwf.int/files/public/cams/products/cams_global_forecast/surface_concentrations/)

Test Approach:

• Clear the qa env forecast database, and run the forecast script to re-populate with data from CAMS
• Use the below query in mongo db to filter all data for 5 days from start date of 17/05/24
• { measurement_date: { $gte: new ISODate('2024-05-17T00:00:00.000Z'), $lte: new ISODate('2024-05-22T00:00:00.000Z') } }
• Export to csv
• Get the forecast for the same time period from ECMWF as CSV
• Run analysis on specific cities for the different available pollutants

Notes:

• PM10 & PM2.5 forecast and database values are very close, usually out by 0.00 – 0.01%
• O3 & NO2 are consistently over 10% out

Conclusion

• Single-level data is a lot more accurate than multi-level, could be problems with the request.