PM concentrations: separate yearly analyses and comparison

Europe/Sofia local time; original averaged export records for 2025, nominal quarter-hour observations for 2024/2026; concentrations in µg/m³.

Full 2024 analysisFull 2025 analysisFull 2026 analysis

Coverage differs substantially between years. The means below describe the available intervals within each local calendar year. Full-year coverage includes every expected interval in that year, including the as-yet unobserved end of 2026. The per-year tables also report coverage through the last observation. Missing periods are left empty. Winter groups January, February and December of the displayed calendar year.

Available-period mean ± sample standard deviation

YearObserved local datesPM₁₀ intervals Full-year coveragePM₁PM₂.₅PM₄PM₁₀
202415 Feb – 31 Dec30,66787.3%9.50 ± 7.6312.11 ± 8.8314.29 ± 11.5921.37 ± 18.90
202501 Jan – 31 Dec34,39597.6%12.08 ± 13.0314.29 ± 14.2616.52 ± 15.5123.10 ± 19.88
202601 Jan – 13 Oct27,19677.6%8.62 ± 6.1410.47 ± 6.8112.38 ± 7.7318.49 ± 11.38

Annual statistics, medians, counts and coverage | Monthly coverage | Seasonal statistics | Monthly statistics

Observed season periods and coverage

Observed winter periods: 2024: 15-29 Feb 2024; Dec 2024 (46/91 dates); 2025: Jan 2025; Feb 2025; Dec 2025 (90/90 dates); 2026: Jan 2026; Feb 2026 (59/90 dates). December of the preceding year is not included. Unequal partial calendar seasons must not be compared as equivalent complete winter seasons. Period ranges below may contain gaps; each observed date has at least one valid PM record, not necessarily a full day of measurements.

YearSeasonObserved local periodsPM recordsDates observed / calendar datesCoverage status
2024Winter15-29 Feb 2024; Dec 20244,41646/91Partial calendar season
2024SpringMar 2024; Apr 2024; May 20248,82892/92All dates represented; check interval coverage
2024SummerJun 2024; Jul 2024; Aug 20248,72192/92All dates represented; check interval coverage
2024AutumnSep 2024; Oct 2024; Nov 20248,70291/91All dates represented; check interval coverage
2025WinterJan 2025; Feb 2025; Dec 20258,61890/90All dates represented; check interval coverage
2025SpringMar 2025; Apr 2025; May 20258,71292/92All dates represented; check interval coverage
2025SummerJun 2025; Jul 2025; 01-28 Aug 20258,38889/92Partial calendar season
2025AutumnSep 2025; Oct 2025; Nov 20258,67791/91All dates represented; check interval coverage
2026WinterJan 2026; Feb 20265,66459/90Partial calendar season
2026SpringMar 2026; 01-05, 07-30 Apr 2026; May 20268,63791/92Partial calendar season
2026SummerJun 2026; Jul 2026; Aug 20268,81592/92All dates represented; check interval coverage
2026AutumnSep 2026; 01-13 Oct 20264,08043/91Partial calendar season

Download season periods, record counts and observed-date coverage.

All-zero record sensitivity

The primary statistics retain measured zeros. The table below shows the effect of excluding fully all-zero analytical intervals as a sensitivity calculation. Mixed zero/nonzero source bins remain included; this is not a complete quality-control filter.

YearAll-zero intervalsPM₁ PM₂.₅PM₄PM₁₀
2024485 (1.58%)9.50 → 9.6612.11 → 12.3114.29 → 14.5221.37 → 21.71
20254 (0.01%)12.08 → 12.0814.29 → 14.2916.52 → 16.5223.10 → 23.10
20260 (0.00%)8.62 → 8.6210.47 → 10.4712.38 → 12.3818.49 → 18.49

Values are original mean → sensitivity mean in µg/m³. Sensitivity counts, means and SDs

Matched calendar comparison

Each pair uses only month/day/hour/quarter-hour slots containing a valid value in both years. Repeated local clock slots on the autumn DST day are averaged within each year before matching. The two means therefore cover the same calendar slots. Matching dates does not guarantee identical weather, weekday composition or instrument condition; differences are descriptive and have no causal interpretation.

YearsMatched PM₁₀ slotsCalendar days MonthsPM₂.₅ (µg/m³)PM₁₀ (µg/m³)
2024 → 202529,768317Feb, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct, Nov, Dec11.99 → 11.94 (-0.4%)20.99 → 20.77 (-1.0%)
2024 → 202622,743240Feb, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct10.97 → 9.87 (-10.0%)20.66 → 18.49 (-10.5%)
2025 → 202626,408282Jan, Feb, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct14.02 → 10.45 (-25.5%)22.72 → 18.38 (-19.1%)

All matched means, medians, SDs and differences

Comparison figures

01_available_period_means
02_monthly_coverage
03_monthly_concentrations
04_seasonal_concentrations
05_diurnal_concentrations
06_daily_time_series_all_years
07_weekday_concentrations
08_weekday_weekend_concentrations
09_pm25_pm10_ratios
10_matched_calendar_means
11_autumn_diurnal_comparison
11_spring_diurnal_comparison
11_summer_diurnal_comparison
11_winter_diurnal_comparison
12_weekday_diurnal_comparison
12_weekend_diurnal_comparison

Calculation definitions

Raw-minute sources use the established UTC-clock interpretation and UTC-aligned 15-minute bins, then convert to local time. The recovered 2025 instrument exports instead retain their original averaged records without reaggregation; missing valid-minute counts are not inferred. Raw bins retain available non-negative source measurements, with their original source counts exported. Monthly, seasonal, hour-of-day and day-type means give each retained interval equal weight. Diurnal means pool the retained averaged records at each hour, and SD in the individual-year graphs describes that interval spread, not uncertainty of the mean. The overall mean is sum of accepted values / number of accepted records, not an unweighted mean of monthly means. Sample SD uses the N-1 denominator. Hour H represents H-1:00 through H-1:59. Ratios are calculated for each interval before averaging.

Instrument identifiers include 03521 and 20746; intercalibration has not been established.