Imagine being able to examine years of sulphur dioxide, nitrogen dioxide, ozone and particulate matter data instead of relying only on one or two days of monitoring. That was the starting point of my CAMS posts and, later, a practical tool for accessing and analysing the data. The purpose is not simply to collect more numbers. It is to understand whether the conditions observed during a campaign fit the longer historical context of the region.
What a short campaign cannot tell us on its own
People working in environmental impact assessment know the value of field monitoring. A study scope or reviewing authority may require actual measurements at the project location. However, some campaigns cover only 24 hours, several days or a week. Those measurements describe that period; they do not, by themselves, describe the rest of the year or the preceding years.
The campaign may coincide with a particular wind season, an unusual event or a period outside the months of highest concentrations. We should therefore ask whether the monitoring window was typical, what happens in other seasons and whether conditions differ between years. A longer record allows those questions to be explored rather than assuming that a short snapshot represents everything.
What CAMS data represent
CAMS is the Copernicus Atmosphere Monitoring Service. Its EAC4 global reanalysis combines observations and modelling to provide a consistent atmospheric composition record from 2003. Estimates are supplied at three-hourly intervals on a 0.75° × 0.75° grid. [1]
This represents a scale of many tens of kilometres, often described approximately as 80 km, with actual dimensions varying by location and direction. It is not a measurement at a factory fence or an individual receptor. Separate European products provide a finer grid of approximately 0.1°, or around 10 km; that European resolution should not be assumed to be available globally. [2]
The questions a historical record can support
Depending on the variable, the record can support examination of monthly and annual patterns, months with higher values, seasonal differences and year-to-year changes that require interpretation. A monitoring period can also be placed alongside the regional record to understand whether it coincides with common or distinctive conditions.
The practical change I proposed was moving beyond a report that says only “air quality was monitored for 24 hours”. A stronger discussion can describe the campaign and add an analysis of several years of regional data. That broader context is useful when the difference between local measurements and a regional grid is kept explicit, rather than asking one cell to resolve details it cannot represent.
Why the work is more than downloading a spreadsheet
Access requires choosing a dataset, variables, area, period and vertical level, then working with scientific files such as GRIB or NetCDF. These are not ready-made Excel worksheets. They organise information across dimensions including space, time and variables, and sometimes vertical levels.
Processing then raises further questions: which grid value should be extracted, how should files be combined, what units are present, and is the variable a mixing ratio or a concentration? Building a time series and calculating a suitable average also require decisions. Converting a mass mixing ratio into a concentration can require appropriate air density; one universal multiplier is not a substitute for understanding the variable.
Why I started building a platform
I wanted to shorten that sequence. The user would select a location on a map or enter coordinates, choose a period and available pollutants, and let the platform create and follow the request to the Atmosphere Data Store, or ADS. It would then manage the processing steps that users would otherwise perform separately.
The beta version I announced covered the available core pollutant data for 2003–2025. Its purpose was to turn scientific files into organised analysis-ready outputs, time series, graphics and reporting material, instead of leaving the user with another NetCDF file and a new list of processing tasks. The project website is cams.aermod.studio. [3]
The journey from request to output
The workflow begins with coordinates, then the time window and pollutants. A request is submitted to ADS and its status is followed. When data arrive, the platform combines the files, prepares the variables, applies the relevant unit conversions and organises the results into usable outputs.
Keeping project coordinates within the workflow makes an analysis easier to reproduce. It does not increase the native spatial resolution. A well-formatted file and a clear graph must not turn a regional estimate into an implied virtual monitoring station at the project site.
Queues, cloud costs and the beta limits
ADS requests operate through a queue. Clicking Download does not guarantee immediate completion: the request must be processed, and timing depends on its size and service conditions. Showing the request status and progress was therefore part of making the experience understandable.
The source data may be free, but servers performing aggregation, conversion, processing and report generation still cost money. In the beta announcement I therefore offered one free trial per user, for one year of data at one location, with an account that needed activation. Those limits allowed me to evaluate real demand, load and operating costs. They described that trial rather than a permanent commitment to identical usage conditions.
Where I saw practical value
I saw applications in impact assessment, baseline studies, academic research and the historical context of air dispersion projects. Comparing a short campaign with a longer record was a particularly practical use: it helps avoid interpreting a limited window without its surrounding history.
Using the data as regulatory background in a quantitative assessment requires additional justification concerning spatial and temporal representativeness, overlapping source contributions and the relevant assessment method. Easy extraction is not, on its own, evidence that every subsequent use is appropriate.
What I wanted to learn from users
I do not present CAMS as a replacement for field measurements where those are required. I see it as another layer of knowledge that adds time and regional context. At the same time, difficult scientific formats should not prevent specialists from exploring useful open evidence.
That is why I asked beta users to report problems and suggestions: are the outputs genuinely useful, are the steps understandable, and what deserves further development? For me, the platform succeeds when it reduces preparation effort while making the source and limitations of the resulting analysis easier to see.
References & original writing
Adapted and expanded from the author’s LinkedIn writing. Technical references are listed below.
