Detecting rice inundation status for water saving and methane emission mitigation measures using Sentinel-1 & ALOS-2/PALSAR-2 Data

Approximately 90% of the world total paddies area and annual output of the rice production are concentrated in monsoon Asia, which has no more land/water resources for further expansion of cultivation. Most rice grows under lowland conditions where currently facing to the fresh water scarcity due to sea-water intrusion accelerated by sea-level rise and land-subsidence, and decelerating freshwater supply by upstream-dam construction. Since the rice production also requires large amount of water (3,000-5,000 L kg^-1 rice), water-saving irrigation practice (e.g., Alternate Wetting and Drying, a.k.a., AWD) is desirable to be implemented in this region to save the water-demand sustainably, and irrigation status need to be evaluated for the decision making on sustainable food security. In addition to the significance of AWD’s role as an adaptation to drought risks, AWD also has a potential to act as an important mitigation-measure by reducing methane emission from paddy soils. This function is very important since rice cropping is responsible for approximately 11% of global anthropogenic CH₄ emissions, and rice has the highest greenhouse gas intensity among the main food crops.

In order to implement AWD in Asian rice paddies as a mitigation-measure based on a carbon pricing scheme, it is important to evaluate the spatial distribution of AWD paddy fields in the target region. For the detection of AWD-fields versus continuously-flooding fields, it is essential to develop method using EO data to detect soil inundation under rice plants at various growth stages.  In this study, ALOS-2/PALSAR-2 and Sentinel-1 data were used to combine the penetration capacity of L-band SAR data with C-band data capacity to monitor rice growth status with their high temporal resolution.

The study was conducted in triple rice cropping systems in the Vietnamese Mekong delta (5 sites: Thot Not in Can Tho city; Chau Thanh, Cho Moi, Thoai Son and Tri Ton in An Giang Province, where AWD field campaign was conducted from 2012 to 2017. EO data consisted of ALOS-2/PALSAR-2 every 14 days in 2017/2018 in An Giang province at high resolution observation mode (3-6m resolution) and ScanSAR observation mode (25-100m resolution) every 42 days over the Mekong delta.

As the result of the classification using the dual-polarization ALOS-2/PALSAR-2 data, soil inundation status could be detected during various rice growth stages. To evaluate rice productivity and GHG emissions from rice fields, we developed a simulation system based on the DeNitrification-Decomposition (DNDC) model which can assimilate PALSAR-2 inundation map and ground observed GHG -flux and rice growth status data on a pixel basis. For spatial extension, rice map, together with rice calendar (sowing date, rice growth status), required as inputs by DNDC are provided by the GeoRice project, based on the use of Sentinel-1 6-day time series. This paper presents the performance of multi-sensor data fusion to realize sustainable agricultural management by mitigating the GHGs emission while maintaining or improving regional fresh water use efficiency for stable food production under climate change pressure.