Recent changes of floods and related impacts in Niger based on the ANADIA Niger flood database

During the last two decades, the sub-Saharan region has experienced unusual floods that have differentially impacted the region. No official and precise data regarding flood damage and impacts on the population are available, and the magnitude of events are not easily evaluated. Most previous studies have investigated this new threat using data derived from local media sources or world disaster databases. The aim of this study was to provide the scientific community and policy makers with an updated and reliable referenced data source concerning floods in Niger between 1998 and 2015, at national, regional and sub-regional scales. Reliable information regarding floods was derived from the national official flood damage database (ANADIA DB) showing their impact on the country. During the investigated period, considerable numbers regarding flood impacts were found (about 4000 settlements and 1.7 million people were affected by floods). The analysis also indicates a sudden increase in flood impacts since 2010. Regions in the south-west (Tillabery, Dosso and Niamey district) are the most affected; however, this kind of risk involves the whole country, and some particularly vulnerable areas have been identified. A data modeling comprehensive framework based on remotely sensed rainfall (climate hazards group infrared precipitation with stations (CHIRPS)) and vegetation index (moderate resolution imagery spectroradiometer normalized difference vegetation index (MODIS NDVI)) datasets data along with census data were used to investigate which variables are most able to explain the recent and sudden Niger flood vulnerability detected at the departmental scale. Only a few statistically significant flood damage models were found (61 out of 297), due essentially to the non-linearity of the increase in damage time series compared to environmental and climatic trends. The population increase is the most significant variable at national level; however, at regional and sub-regional scales, different patterns provided evidence to identify local triggers for vulnerability.