Abstract

Recent conflict along the border of Bangladesh and Myanmar has amplified a food security crisis and access to the region remains challenging. Moderate-resolution satellite remote sensing offers an approach to complement more traditional food insecurity hot spot assessment across Rakhine, Myanmar; however, conflict creates unique signals that are not agroclimatologically driven and need to be considered. Time series radar and optical data cubes were built and used to assess for deviations across space and time for rice paddy production areas based on established techniques. Ultimately, the Sentinel-1 radar was more helpful compared to fused Landsat-7 and -8 and Sentinel-2 data cubes that were substantially impacted by cloud cover during key growth stages. Anecdotal reporting, very high resolution (VHR) imagery, and expert knowledge were used to support operational analyses routines in an attempt to characterize rice into failed, abandoned, and cultivated classes across 2016 to 2018 seasons. Accuracy assessment using co-timed VHR showed overall accuracy (%) of 86.5, 87.5, and 91.0 for 2016, 2017, and 2018, respectively. Nearly one-third of rice production was characterized as failed or abandoned in any given year. Qualitative analyses showed paddy failure was often adjacent to conflict events. The moderate-resolution imagery and automated routines offer complementing metrics that can be used to help guide food security assessments. In regions where climate change, migration, and conflict coincide, decision support tools will need to evolve and continue to integrate human perspectives. View Full-Text