Swaraj1313/Big-Data-Post-Cyclone-Idai-Recovery-Mozambique

This project applies satellite time-series analysis to assess the long-term impact of Cyclone Idai on cropland productivity across central and northern Mozambique. # Mozambique-Idai-Cropland ### Cropland Vulnerability and Post-Disaster Agricultural Recovery Assessment #### Mapping the Impact of Cyclone Idai on Crop Production Capacity, 2017–2024 --- ## Overview This project applies satellite time-series analysis to assess the long-term impact of Cyclone Idai on cropland productivity across central and northern Mozambique. Using Sentinel-2 NDVI composites spanning eight growing seasons (2017–2024) and a fixed cropland mask derived from ESA WorldCover 2021, the analysis quantifies crop production recovery at both pixel and district level — producing sub-national evidence on agricultural resilience and persistent food security risk that conventional survey methods cannot provide at this spatial resolution. Cyclone Idai made landfall near Beira on 15 March 2019, generating storm surges, prolonged inland flooding along the Buzi and Pungwe river systems, and widespread cropland inundation across Sofala, Manica, Zambezia, and Tete provinces. The storm is one of the most destructive tropical cyclones on record in the Southern Hemisphere. Its agricultural impact — both immediate and multi-year — has direct implications for food security across a population heavily dependent on subsistence rain-fed agriculture. The analytical framework follows the established remote sensing approach for post-disaster agricultural assessment: a pre-event NDVI baseline is established from two full growing seasons, and recovery is tracked season by season against that baseline. Districts and pixels that fail to reach 85% of baseline productivity by the 2023–24 growing season are classified as exhibiting persistent degradation, representing cropland whose production capacity has not been restored five years after the event. --- ## Why This Is a Big Data Problem Mapping cropland recovery across four provinces at 20-metre resolution, season by season for eight years, is not something you can do on a laptop. The numbers give a sense of the scale: rough …