Hawa Ka Hisab: InvestigationsDiwali night, 2025
Delhi PM2.5 monitoring coverage
Published by Ajay Maken, Member of Parliament, Rajya Sabha. Audited on 18 October 2026.
The saved hourly record thinned overnight. Delhi needs an explanation for each missing station-hour.
12 stations at 02:00 on 21 October, compared with 40 one week earlier.
Red: 20 to 21 October 2025. Blue: 13 to 14 October 2025. All hourly labels are IST.
Delhi needs a complete hourly record when pollution is a concern. On the night of 20 to 21 October 2025, the saved PM2.5 archive contained readings from fewer stations than it did a week earlier.
At 02:00 on 21 October, the archive contained numeric PM2.5 readings from 12 stations. It had readings from 40 stations at 21:00 the previous evening. The matched 02:00 timestamp on 14 October had readings from 40 stations.
The seven consecutive hourly timestamps from 23:00 through 05:00 each had fewer than 30 stations with a numeric observation. The same seven timestamps one week earlier each had 40 stations.
This is a finding about the archived record. The absence of a reading cannot by itself establish whether an instrument stopped, whether transmission failed, or whether the archive omitted a value. It cannot reveal the missing concentration or establish how missing readings changed a city average.
The comparison covers 24 hourly labels from noon on 20 October through 11:00 on 21 October, beside the matched 13 to 14 October window. The timestamps are IST. This is an hourly coverage comparison, not a daily AQI calculation or an estimate of emissions.
Each point counts distinct CPCB site identifiers, rather than station-name spellings. Raw numeric coverage and the existing PM2.5 analysis filter, above 0 and below 1000 micrograms per cubic metre, give identical counts in these selected windows. No daily mean or daily station-day completeness threshold is used to count the hourly observations.
The appropriate next step is to reconcile the missing station-hours with the primary station records and publish the reason for each absence. The archive alone cannot supply that explanation.
All 48 hourly counts
| Diwali timestamp, IST | Stations | Comparison timestamp, IST | Stations |
|---|
| 2025-10-20 12:00 | 40 | 2025-10-13 12:00 | 38 |
|---|
| 2025-10-20 13:00 | 40 | 2025-10-13 13:00 | 40 |
|---|
| 2025-10-20 14:00 | 40 | 2025-10-13 14:00 | 39 |
|---|
| 2025-10-20 15:00 | 39 | 2025-10-13 15:00 | 39 |
|---|
| 2025-10-20 16:00 | 39 | 2025-10-13 16:00 | 40 |
|---|
| 2025-10-20 17:00 | 40 | 2025-10-13 17:00 | 40 |
|---|
| 2025-10-20 18:00 | 40 | 2025-10-13 18:00 | 40 |
|---|
| 2025-10-20 19:00 | 40 | 2025-10-13 19:00 | 40 |
|---|
| 2025-10-20 20:00 | 40 | 2025-10-13 20:00 | 40 |
|---|
| 2025-10-20 21:00 | 40 | 2025-10-13 21:00 | 40 |
|---|
| 2025-10-20 22:00 | 35 | 2025-10-13 22:00 | 40 |
|---|
| 2025-10-20 23:00 | 28 | 2025-10-13 23:00 | 40 |
|---|
| 2025-10-21 00:00 | 19 | 2025-10-14 00:00 | 40 |
|---|
| 2025-10-21 01:00 | 15 | 2025-10-14 01:00 | 40 |
|---|
| 2025-10-21 02:00 | 12 | 2025-10-14 02:00 | 40 |
|---|
| 2025-10-21 03:00 | 16 | 2025-10-14 03:00 | 40 |
|---|
| 2025-10-21 04:00 | 25 | 2025-10-14 04:00 | 40 |
|---|
| 2025-10-21 05:00 | 22 | 2025-10-14 05:00 | 40 |
|---|
| 2025-10-21 06:00 | 32 | 2025-10-14 06:00 | 40 |
|---|
| 2025-10-21 07:00 | 35 | 2025-10-14 07:00 | 40 |
|---|
| 2025-10-21 08:00 | 38 | 2025-10-14 08:00 | 39 |
|---|
| 2025-10-21 09:00 | 38 | 2025-10-14 09:00 | 40 |
|---|
| 2025-10-21 10:00 | 38 | 2025-10-14 10:00 | 40 |
|---|
| 2025-10-21 11:00 | 40 | 2025-10-14 11:00 | 39 |
|---|
The windows contain 791 and 954 station-hours respectively. A station-hour means one distinct station with a numeric observation at one hourly label.
Read the two-page PDF | Download the checked counts (CSV)
Source raw archive SHA256:
4b84a34fb001539199e36b8783d934652f6cd620d801a1d7f96f3b4f32991c60
CREA data API and documentation. CREA has no involvement in this analysis.