There have been some rapid releases of the {ecmwfr} R package of late (i.e. going from v2.0.0 to v2.0.2). This is in part due to the migration to the new ECMWF API framework. Michel Bobbia, a statistician at Atmo Normandie, recently highlighted some issues downloading European CAMS data which weren’t apparent through our integrated testing. Atmo Normandie provides air quality indices across Normandy in northwestern France. Michel uses EU CAMS, and I quote,
“… forecasts to estimate ATMO index, based on O3, NO2, PM10, PM2.5 and SO2 daily levels. We get 7 model forecasts, and calculate a mixing model with them. So we have 8 models to predict our Air Quality index over our region for today and tomorrow, which leads in the maps in our website. I mainly code in R because I’m a statistician. The code is executed every day, gets the ‘ensemble’ model results and saves them in a database. The it calculates the index and produces a map. We also have a tool for comparing models with our measurements …”
This is a reminder that community based infrastructure only works if members highlight these issues and try to document them. This bug is now fixed in the most recent v2.0.2 version of {ecmwfr} on github which can be installed using:
remotes::install_github("bluegreen-labs/ecmwfr")
library("ecmwfr")
The package has been submitted to CRAN and will probably be released somewhere next week.
I would like to thank Michel for bringing this bug to our attention, and indirectly highlighting how the {ecmwfr} R package is used within an operational setting to inform the public on air quality.
