How many stations have valid data on Romania's air?
In 2021, 134 stations in Romania's national network had valid nitrogen dioxide data for at least three quarters of the year's hours. In 2022 there were 61, in 2023, 29, and in 2024, 25. The counts come from the data Romania sent, verified, to the European Environment Agency. Three quarters is the threshold the Agency uses in its statistics; the legal one is stricter. A month without data means this record holds no valid values, not necessarily that the instrument was off. Pick a county and see, month by month, when its stations have valid data and when they do not.
Each row is a station, each cell is a month#
A lit cell is a month with valid measurements; the paler it is, the fewer hours or days were covered. The dark areas are months for which the European record holds no valid value from that station. Stations are grouped by county, in alphabetical order.
The station data did not load. Reload the page.
Stations with valid nitrogen dioxide data for at least three quarters of the year, 2013–2025. Out of 157 stations that ever reported this gas.
The 2022–2024 gap is not the first. In 2013–2016, too, the record has at most 38 stations a year with valid nitrogen dioxide data; this page cannot say whether the stations were not measuring then or the data did not reach this reporting flow. Across all 13 years the best year was 2021, with 134 stations. 112 of the 157 stations have at least one pause longer than two years between two hours with valid values; at half of the stations the longest pause is at least 899 days.
Three quarters is the most lenient threshold in use: the European Environment Agency includes in its annual statistics stations with more than 75% valid data. The EU directive in force in these years asks for 90%, not counting losses from calibration and maintenance, and the new directive asks for 85%. The table gives the count at each threshold.
| Pollutant and threshold | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| NO₂ · ≥ 75% | 36 | 20 | 32 | 38 | 106 | 108 | 81 | 105 | 134 | 61 | 29 | 25 | 115 |
| NO₂ · ≥ 85% | 25 | 14 | 11 | 21 | 92 | 99 | 74 | 89 | 122 | 60 | 27 | 21 | 91 |
| NO₂ · ≥ 90% | 19 | 9 | 4 | 8 | 70 | 77 | 62 | 67 | 100 | 52 | 19 | 15 | 77 |
| PM10 · ≥ 75% | 53 | 43 | 42 | 28 | 69 | 88 | 64 | 80 | 111 | 68 | 67 | 77 | 142 |
| PM10 · ≥ 90% | 20 | 15 | 18 | 19 | 34 | 42 | 34 | 52 | 77 | 49 | 53 | 51 | 112 |
| PM2.5 · ≥ 75% | 9 | 6 | 5 | 4 | 10 | 13 | 6 | 10 | 23 | 14 | 30 | 56 | 64 |
What could no longer be known#
Without enough measurements nobody can say a limit was met. In 2021, five stations recorded more than 35 days above 50 µg/m³ of PM10 particles, which breaches the legal daily limit: Craiova, Rovinari (Gorj county), Iași, Ungheni (Iași county), Târgu Mureș. For 2023, none of them has a valid PM10 year on record. Three of them (Craiova, Iași, Târgu Mureș) have no value at all for that year. The other two have incomplete years in which the limit still appears breached: 69 days above the threshold at Rovinari, with 64% of days covered and 86 days above the threshold at Ungheni, with 68% of days covered.
For nitrogen dioxide, 62 stations had a 2021 annual mean above 20 µg/m³, the limit set for 2030. Of these, 17 have a valid year in 2023 and 14 in 2024.
In 2023, 56 of the 110 towns and communes with stations had no station with a valid year for any of the three pollutants followed here. About 1.95 million people lived in them at the 2021 census. In 2021 there were 21 such places. For nitrogen dioxide, 26 of Romania's 42 counties have no valid year in 2023 and 28 in 2024, against one in 2021.
What data kept coming
PM10 particles are also measured differently from gases: a filter collects dust for a whole day and is weighed in a laboratory. These daily samples continued at many stations. With them, the number of stations with a valid PM10 year fell from 111 in 2021 to 68 in 2022 and 67 in 2023, then rose to 77 in 2024.
In Bucharest the number of stations with a valid nitrogen dioxide year barely changed: 6 in 2021, 5 in 2023, 7 in 2025. The stations still on record in 2023 are mostly those in the large cities. Of the 134 stations with a valid 2021, 24 also have a valid 2023; their 2021 nitrogen dioxide mean was 26.1 µg/m³, against 19.6 µg/m³ at the 110 that are missing. A national average over the remaining stations therefore cannot be compared with the 2021 one.
Are the data missing just when the air is dirtier? Two tests#
If the missing hours were the most polluted ones, annual means would come out too low even at stations that pass the threshold. This was checked in two ways, with thresholds fixed before counting.
The last hour before a pause not confirmed
For 9,801 pauses of at least six hours, the last measured value was ranked among the same station's values in the same month at the same hour of day. If data went missing after unusually polluted hours, that last value would sit, on average, well above the middle of the ranking. It sits at 48.1% (the middle is 50%; the threshold set in advance was 55%).
The neighbouring station, during the pause a signal for PM10 not for NO₂
When a station has no data for between six hours and 30 days, what does another station within 15 km measure? For nitrogen dioxide, neighbours read 98% of their level in the rest of the same month, so no increase. For automatically measured PM10, neighbours read 18% more (95% confidence interval: 7% to 30%), and the ratio is above 1 in each of the years 2022–2025. The threshold set in advance was 10%. For PM2.5 the result is inconclusive.
The test does not show why. One possible explanation, not checked here, is that the weather that keeps dust near the ground, such as fog and cold still air, also disrupts the instruments. The practical consequence stands: for PM10, an annual mean computed from an automatic series with pauses risks being lower than the real one.
What comes next: the 2030 limits#
In 2025 the data came back: 115 stations have a valid nitrogen dioxide year. No station with a valid year exceeds today's permitted annual means; 12 stations do breach the daily PM10 limit, with more than 35 days above 50 µg/m³. Directive (EU) 2024/2881 sets lower limits from 1 January 2030: 20 µg/m³ for nitrogen dioxide, 20 for PM10 and 10 for PM2.5. Against those, in 2025 the stations above the limit were 43 of 115 for nitrogen dioxide, 83 of 142 for PM10 and 59 of 64 for PM2.5. Member states must transpose the directive by 11 December 2026.
On 11 December 2025 the European Commission decided to refer Romania to the Court of Justice of the EU, because its national monitoring network does not meet the EU directives' requirements on the number of sampling points and on data quality; the procedure began in 2017. The European Environment Agency estimates that 14,065 deaths in Romania in 2023 were attributable to PM2.5 particles.
Sources and limits#
- Source. Verified hourly and daily values (the E1a data flow) for nitrogen dioxide, PM10 and PM2.5, reported by Romania and published by the European Environment Agency, downloaded on 12 November 2026; 197 stations, 2013–2025. EEA data may be reused with the source named.
- What “valid year” means. For nitrogen dioxide: valid hourly values for at least 75% of the year's hours. For particles: at least 75% of days, a day being covered by a valid daily sample or by at least 18 valid hourly values from an automatic instrument (the day's value is the daily sample or, without one, the mean of the hours). Negative values flagged as valid were not counted.
- The record is not the instrument. A dark month means the verified record sent to the Agency holds no valid values. For 2022 and 2023 the record contains almost only whole station-years or nothing (only 5 and 6 stations have partial years), so a station may have measured for part of a year without those values being reported as verified. In 2024, 62 stations have at most a quarter of the year on record. The raw data on calitateaer.ro were not compared here.
- Why the data are missing. This page counts; it does not establish the cause. In 2023 the Romanian press reported a member of parliament's statement that from the end of 2021 there were no signed maintenance contracts for the monitoring stations (Jurnalul, 6 December 2023; Spotmedia, 6 April 2024). No public explanation from the Ministry of Environment for these years was found.
- Places and residents. Each station was placed in a town or commune by the coordinates in the EEA metadata. The population is that of the whole locality at the 2021 census; it does not say how many people breathe the air a station measures.
- The tests. Thresholds and rules were written before the first count. The first hypothesis was not confirmed and is shown as such.