Understanding Punjab through Earth’s changing reflectivity
An independent NASA Space Apps Challenge 2026 research project.
One landscape.
Many connected systems.
Earth System Trend Detective investigates how aerosol optical depth, UV aerosol absorption and surface albedo change across Punjab, Pakistan, using only supplied observations and clearly marking variables whose data are not loaded.
Built for the NASA Space Apps Challenge 2026 challenge, “Be an Earth System Trend Detective.” Independent research—not developed, operated, or endorsed by NASA.
Our research question
How have aerosol optical depth and surface albedo changed across Punjab, how do the loaded records relate, and what does the supplied UV Aerosol Index reveal about absorbing haze?
Methodology & sources
About this data
| Variable | Source | Period | Resolution | Valid months | Last updated |
|---|---|---|---|---|---|
| Surface albedo | NASA Giovanni / MERRA-2 | 2005-01 to 2025-01 | Monthly area average; 0.5° × 0.625° source grid | 241 | 2026-11-15 |
| Aerosol optical depth | NASA Giovanni / MODIS Aqua | 2005-01 to 2025-12 | Monthly area average; 1° × 1° product grid | 252 | 2026-11-15 |
| UV Aerosol Index | NASA Giovanni / Aura OMI OMAERUVd v003 | 2005-01 to 2025-12 | Daily 1° source grid; cleaned monthly area averages | 252 | 2026-11-15 |
OMI OMAERUVd UV aerosol index: 252 supplied cleaned monthly means, 2005–2025. UVAI trend statistics and confidence bounds are supplied findings, not recalculated. The OMI row anomaly and artifact-removal sensitivity limit whole-year trend claims.
How we cleaned the data
450 days were missing (NASA fill values). We removed 239 more days (3%) that were sensor artifacts: values up to 34, jumping on alternate days in a repeating pattern linked to OMI's known 'row anomaly'. We used a Hampel filter (more than 4 robust standard deviations from the 15-day median) plus a cap at 5. Every month keeps at least 18 valid days.
| Cleaning sensitivity | Whole-year p |
|---|---|
| None | 0.0003 |
| Cap only | 0.050 |
| Hampel 15-day (k=3) | 0.074 |
| Hampel 31-day (k=3) | 0.119 |
The winter rise is significant under every method; the whole-year trend is not robust.
Sensitivity runs use k=3; the supplied primary clean series uses k=4. Findings are supplied statistics, not recalculated here.
Download daily raw & clean dataA study extent, not a boundary
The working rectangle is 70°E–75.5°E, 28°N–34°N. It includes neighbouring areas. The actual MERRA-2 extraction is 69.375°E–75°E, 28°N–34°N.
The CSV is MERRA-2
The brief and research dossier discuss an earlier North American NLDAS sample. The currently attached CSV identifies M2TMNXRAD 5.12.4: global monthly MERRA-2 surface albedo, 0.5 × 0.625°.
Respect incomplete years
Surface albedo: 2005-01-01–2025-01-01 (241 observations). Aerosol optical depth: 2005-01-01–2025-12-01 (252 observations). UV Aerosol Index: 2005-01-01–2025-12-01 (252 observations). 2025 albedo is partial.
From observations to interpretation
Source register
Original research files
The dossier audits a different NLDAS export; do not transfer its exclusion or findings onto the attached MERRA-2 series.