Global_Flood_Intercomparison_Project

The Global Flood Intercomparison project ran from 2023 to 2025 under the NASA/USAID SERVIR program. The project aimed to facilitate access to evaluate the performance of satellite-derived surface water products. produced by NASA, NOAA, and ESA. The project had two main deliverables, one being the Flood Mapping Intercomparison (or FMI) software & repository, which is linked in the "Related Works" section. This repository contains more information regarding each of the flood products. The flood products included in this study were the following:

• DSWx-HLS: Dynamic Surface Water eXtent from Harmonized Landsat Sentinel 
• DSWx-S1: Dynamic Surface Water eXtent from Sentinel-1
• GFM: Global Flood Monitor 
• HYDRAFloods: HYDrologic Remote Sensing Analysis Tool for Floods 
• HYDROSAR
• MCDWD: Global Flood Product -- MODIS Combined Water Detection 
• VFM: VIIRS Flood Mapper

More details regarding these flood products can be found in the FMI repository linked in related works. The second deliverable is the contents of this zenodo, which is a collection of over 10,000 points stating the presence or lack thereof of surface water during flood events. These points were used to assess the relative accuracy of surface water products, an analysis of which will be released in a forthcoming scientific publication. These points correlate to five flood events for which there were synchronous overpasses of the satellites from which the aforementioned flood products are derived.

• Arkansas: 2025/04/09
• Cambodia: 2024/10/01
• Chad: 2024/09/26
• Florida: 2024/10/15
• Pakistan: 2023/06/16 

These flood events were sourced from the EM-DAT database. Querying this database for these locations and dates will yield more details about these specific flood events. For each of these flood events there are the following items in this database. The naming convention is specified below: 

• reference data
  • naming convention: reference_data_<flood_event>.csv
  • description: This is the point data indicating the presence of surface water or lack thereof for the flood event in question. This was determined via ocular interpretation of Planet Labs imagery obtained on the same date as the flood event in question. 
• harmonized flood products 
  • naming convention 
    • DSWx-HLS:  dswxhls_<flood_event>.tif
    • DSWx-S1: dswxs1_<flood_event>.tif
    • GFM: gfm_<flood_event>.tif
    • HYDRAFloods: hydrafloods_<flood_event>.tif
    • HYDROSAR: hydrosar_<flood_event>.tif
    • MCDWD: mcdwd_<flood_event>.tif
    • VFM: vfm_<flood_event>.tif
  • description: each of the flood products are harmonized to a schema of 0s and 1s, with 1s representing water, 0s representing nonwater, and 2 representing invalid data 
• areas of interest 
  • naming convention: aoi_<flood_event>.geojson
  • description: The area for which all flood products had a common overlap.