Development of satellite-based surface methane flux model for major agro-ecosystems using energy balance diagnostics

Present study was carried out to develop multiple linear regression (MLR) model of surface CH4 flux emission from monthly atmospheric clearness index (8 km), day-night land surface temperature (LST) at 1 km and surface soil moisture (25 km) from Kalpana-1, MODIS TERRA and GCOM-W1 satellites, respectively. All these products were aggregated to GOSAT level-4A product resolution. 2° × 2° grids representing homogeneous agro-ecosystems were used to draw data samples. Initial results showed that methane flux (from GOSAT) produced significant coefficient of determination (R2 = 0.84) with tri-variate (LST, surface soil moisture and atmospheric transmissivity) as compared to bi-variate (LST-soil moisture, LST-atmospheric transmissivity, soil moisture-atmospheric transmissivity) MLR models. These have been utilised for predicting surface methane flux for monthly scale. Validation of predicted methane flux with actual GOSAT methane flux was carried out and RMSE of 4.2–15.9% was obtained using variance-based bias correction. All these scaling models may be utilised to predict CH4 flux at regional level using high-resolution LST from thermal remote sensing and soil moisture from Synthetic Aperture Radar.