Sentinel-1 C-Band SAR for Rice Flood Mapping

This page is written for practitioners and technically literate buyers. It states what we measure, what we infer, and where C-band SAR fails.

Sensor specifics: Sentinel-1 IW GRDH

Parameter	Value	Notes
Mission	Sentinel-1 (Copernicus)	C-band SAR constellation
Frequency	~5.405 GHz (C-band)	ESA mission specification
Mode / product (typical for land)	IW GRDH	Interferometric Wide swath, Ground Range Detected High-res
Pixel spacing	~10 m	IW GRDH; not the same as independent information content
Revisit	~6–12 days	Depends on dual-satellite availability and latitude
Polarisation	VV / VH (common dual-pol over land)	See ESA user guide for acquisition plans
Archive	Continuous public since 2014	Baseline history without project-funded tasking

Exact product choices (calibration, speckle filter, incidence-angle normalisation) belong in a methods preprint. Treat the table as the sensor envelope, not a processing recipe.

Why the historical archive matters commercially

Because Sentinel-1 has imaged monsoonal Asia continuously since 2014, a multi-year flooding-regime history already exists for most paddies without anyone having commissioned it for carbon markets. That archive supports baseline construction and additionality arguments that survey programmes cannot reconstruct after the fact. It is one of the few public goods in rice methane MRV that does not need to be purchased plot by plot.

Sentinel-1 vs Sentinel-2 for rice monitoring

Sentinel-2 optical imagery needs a clear sky. Indian kharif rice coincides with monsoon cloud. Regional cloud-fraction studies commonly show large fractions of growing-season days with cloud cover over major rice belts—often high enough that optical time series lose weeks of observations precisely when drying cycles matter. (Cite the specific cloud-climatology paper or IMD/MODIS analysis you use in the validation report; do not invent a single national percentage here.)

Under cloud, optical flood classification collapses. C-band SAR continues because it illuminates the scene. Side-by-side real S1/S2 scenes for a validation district will replace the conceptual comparison once released on Research. Until then, treat any animated demo as illustration, not evidence.

Where SAR is hard

C-band rice flood mapping is not a solved classification toy. Failure modes we treat as first-class:

Canopy attenuation. As biomass builds, the canopy increasingly interacts with C-band energy; late-season flooded vs drained contrast can weaken.
Speckle. Multiplicative noise requires filtering or multi-temporal methods that trade spatial detail for stability.
Incidence-angle effects. Backscatter varies across the swath; uncorrected angle trends impersonate wetness change.
Wind roughening. Wind on open water increases backscatter and can brighten flooded surfaces toward drained-like returns.
Bright confusers. Drained paddy is not the only bright surface; bare soil, infrastructure, and some crop stages confuse single-date thresholds.
Sub-pixel mixing at ~1 ha. Ten-metre pixels still mix within smallholder parcels, dikes, and canals; field-boundary error propagates into flooded-area fractions.

Handling these is why SAR outputs feed sparse sensor sampling and why issuance depends on uncertainty quantification—not a single backscatter threshold.

Method boundary: what we infer vs what we measure

Measured (remotely): multi-temporal C-band backscatter (and derived flooded/drained classifications) under cloud-penetrating observation.
Measured (in situ): water-level and calibration samples on a designed subset of plots.
Inferred: methane flux via a calibrated model; credit tonnes after uncertainty treatment.

Hub: how verification works. Agronomy context: Alternate Wetting and Drying.

References

Sensor parameters cite ESA documentation. The peer-reviewed entries below are representative Sentinel-1 rice/flood mapping literature; the validation report will cite the exact subset used in our pipeline.