Coverage
Algae data availability, 1997 → 2026
Thirteen sensors, drawn from the database. The 2015 line is the one that matters — before it you have Landsat at 16 days and MODIS at 250 m; after it you add Sentinel-2 at 5 days and Sentinel-3 at 1–2 days. Any model spanning that boundary must bias-correct across sensors, or the step change becomes a spurious trend.
Sensor
2000
2005
2010
2015
2020
2025
Where it applies
Landsat 5 TM
30 m · USGS
16-day
ESA Ocean Colour CCI v6
4000 m · ESA / Plymouth Marine Lab
MONTHLY
SeaWiFS L3 ocean colour
9000 m · NASA OB.DAAC
daily to monthly
Landsat 7 ETM+
30 m · USGS
16-day, SLC gap after 2003
MODIS Terra
250 m · NASA
DAILY
MODIS Aqua L3
4600 m · NASA OB.DAAC
daily to 8-day
VIIRS SNPP / NOAA-20
4000 m · NOAA CoastWatch
DAILY
Landsat 8 OLI
30 m · USGS
16-day
Sentinel-1 SAR
10 m · ESA Copernicus
6-12 day, cloud-proof
Sentinel-2 MSI
20 m · ESA Copernicus
5-day
Sentinel-3 OLCI
300 m · ESA Copernicus
1-2 day, 620 nm phycocyanin
Landsat 9 OLI-2
30 m · USGS
16-day
PACE OCI hyperspectral
1200 m · NASA
1-2 day
inland high-resolution
estuary 300 m
Landsat backfill
MODIS daily
ocean colour
SAR / hyperspectral
What each sensor actually delivers
| Sensor | Provider | Resolution | Revisit | Start | End | Datasets | Docs |
|---|---|---|---|---|---|---|---|
| Landsat 5 TM | USGS | 30 m |
16 d | 1997 | 2011 | provider ↗ | |
| ESA Ocean Colour CCI v6 | ESA / Plymouth Marine Lab | 4000 m |
30 d | 1997 | 2026 | AL01 AL11 | provider ↗ |
| SeaWiFS L3 ocean colour | NASA OB.DAAC | 9000 m |
1 d | 1997 | 2010 | AL07 | provider ↗ |
| Landsat 7 ETM+ | USGS | 30 m |
16 d | 1999 | active | AL20 | provider ↗ |
| MODIS Terra | NASA | 250 m |
1 d | 2000 | active | AL13 AL16 | provider ↗ |
| MODIS Aqua L3 | NASA OB.DAAC | 4600 m |
1 d | 2002 | active | AL06 | provider ↗ |
| VIIRS SNPP / NOAA-20 | NOAA CoastWatch | 4000 m |
1 d | 2012 | active | AL08 | provider ↗ |
| Landsat 8 OLI | USGS | 30 m |
16 d | 2013 | active | AL05 | provider ↗ |
| Sentinel-1 SAR | ESA Copernicus | 10 m |
9 d | 2014 | active | provider ↗ | |
| Sentinel-2 MSI | ESA Copernicus | 20 m |
5 d | 2015 | active | AL03 AL18 AL19 | provider ↗ |
| Sentinel-3 OLCI | ESA Copernicus | 300 m |
1.5 d | 2016 | active | AL04 | provider ↗ |
| Landsat 9 OLI-2 | USGS | 30 m |
16 d | 2021 | active | provider ↗ | |
| PACE OCI hyperspectral | NASA | 1200 m |
1.5 d | 2024 | active | AL15 | provider ↗ |
Cadence — what is actually defensible
| Cadence | Chakaria | Matlab | Dhaka | Why |
|---|---|---|---|---|
| Daily | 2017 onward only | 2017 onward only | not defensible | Dhaka would be ~78% filled even after 2017; before 2015 it is ~95% filled. |
| Weekly | 2015 onward | 2000 onward | 2015 onward | Matlab gets there earliest because MODIS 250 m works on the Meghna from 2000. |
| 10-day | throughout | throughout | 2013 onward | The safest common cadence across all three sites. |
| Monthly | throughout | throughout | throughout | The only cadence every site supports for the full 2000–2026 window. Use it for the cross-site gradient test. |
Cadence is not a preference. It is set by revisit interval and cloud. These verdicts come from the
observed-day fractions that
11_algae_daily_fill.py measures and writes to
algae_daily_fill_report.csv — re-check them against your own AOIs before committing.
Lag grid. 0, 1, 2, 3, 4, 6, 8, 10, 12, 16, 20, 26, 39, 52 ISO weeks. Pre-register the lag you expect for each variable:
34 variables × 14 lags × 3 sites is 1,428 cross-correlations, and some will clear p<0.05 by chance alone.
See Variables for the expected lag per variable.