A GISTDA-led framework combines Sentinel-1 and ALOS-2 radar with IPCC accounting to detect crop timing, classify water regimes, and estimate methane from irrigated rice in Thailand's central plain. Mean absolute error was 18.5 kg CH4/ha, but structural uncertainty in the IPCC Tier 1 emission factor exceeded all algorithmic errors combined.
Key findings
- Mean absolute errors for planting and harvest dates were 6.1±1.4 and 8.3±1.7 days, with a 97.0%±2.7% operational detection rate. Water-regime balanced accuracy ranged from 0.59 to 0.89. Full-pipeline error was 18.5±4.5 kg CH4/ha (21.4% of the mean ground-based calculation), with mean bias of 3.5±5.8 kg CH4/ha.
Why this matters globally
The spatially explicit IPCC-compatible approach could support monitoring, reporting, and verification for rice cultivation and water-management interventions, but national inventories require field calibration and country-specific emission factors.
Thai researcher contribution
GISTDA researchers led a framework tailored to Thai rice landscapes with JAXA and French collaborators, connecting Earth observation technology to agricultural carbon accounting.
Limitations to consider
The study focuses on irrigated paddies in Thailand's central plain; some water-stage classifiers achieved balanced accuracy of only 0.59, and the ground reference was a calculation rather than direct flux measurement at every field. The IPCC Tier 1 structural range (-32% to +48% of default) exceeded algorithmic error, so this is not direct emissions measurement or a certified MRV system.
Verify the original sources
Remote SensingRead the original article↗DOI: 10.3390/rs18132194