Dataset on the expansion and consolidation of flooded settlements in the Dosso Region, Niger

Flood risk reduction at the local scale requires knowledge of the settlements which are most exposed to floods, and those where the existing measures are insufficient to handle the threats. The knowledge on spatial dynamics of the flooded human settlements is limited, especially that of the smaller ones, such as the settlements in the sub-Saharan Africa. The dataset on 122 flooded settlements in the Dosso Region (Niger) offers information on: the built-up area and the number of buildings with corrugated iron roofs in 2004, 2012, and 2019 (average dates), the type of human settlements (city, rural town, village, or hamlet), the flood dates and the number of buildings collapsed between 2011 and 2019. The data on the built-up area and the number of buildings with corrugated iron roofs were extracted by visual photointerpretation from very high-resolution images accessible through Google Earth Pro. The information on the settlement category was obtained from the Human Settlements National Directory (French acronym, ReNaLoc) published by the National Institute of Statistics of Niger. The dates of floods and the data on the number of collapsed buildings were obtained from the open access national database on flooding, known by the French acronym, BDINA. These data can be reused to build a geodatabase for flood risk reduction and to draft the municipal and regional development plans. Their potential reuse allows for the identification of settlements undergoing the most rapid physical expansion, built-up area in a flood-prone zone, and settlements that require protection and flood risk reduction policies. Additionally, the dataset can also be used to verify the accuracy of the built-up area obtained from the satellite images with coarse resolution and for comparisons with other regions in Niger and in sub-Saharan Africa.

a flood-prone zone, and settlements that require protection and flood risk reduction policies. Additionally, the dataset can also be used to verify the accuracy of the built-up area obtained from the satellite images with coarse resolution and for comparisons with other regions in Niger and in sub-Saharan Africa.
© 2022 The Author(s

Value of the Data
• The built-up expansion and consolidation dataset presents an unprecedented overview of the cities, rural towns, villages, and hamlets damaged by floods between 2011 and 2019 in the Dosso Region. The accuracy of the built-up area is higher than those provided by coarse resolution satellite imagery. The dataset is useful for targeting settlements with risk reduction policies and for identifying the expansion areas. • Local governments (municipalities), the regional government, Official development aid, NGOs, and researchers are the main beneficiaries of the data. • The data can be used (i) to identify the most rapidly expanding settlements [1] , (ii) to verify the accuracy of the built-up area obtained from the coarse resolution images, (iii) to identify the extent of the built-up area under the risk of flooding; (iv) to establish the risk reduction measures; and (v) to determine the extent and the speed of consolidation of the buildings with corrugated iron roofs. • The value of these data lies: (i) in their detail, as they are extracted by visual photointerpretation of VHR satellite images by operators experienced with the observed context; and (ii) in their high representativeness, as they cover 40% of the flooded settlements in the Dosso region between 2011 and 2019.

Data Description
The dataset includes five figures and three tables. Fig. 1 shows the location of the 122 human settlements inundated between 2011 and 2019, covered by very high-resolution (VHR) satellite images on three dates between 2001 and 2020, accessible via GE Pro. Fig. 2 shows the dates on which the VHR satellite images of the 122 settlements are freely available on GE Pro. The tables list each of the 122 flooded settlements, ranked alphabetically, provided with a code that allows them to be identified in Fig. 1 , the municipality and the settlement category (hamlet, village, rural town, or city) to which they belong. This is followed by data on built-up area expressed in square kilometres, the year being referred ( Table 1 ), the number of buildings with corrugated iron roofs on three dates (the years are the same as those reported in ( Table 1 ), and the total number of buildings in 2012. The last two fields of information are limited to only 49 settlements ( Table 2 ).
The last table shows the number of buildings that collapsed as a result of flooding in the period 2011-19 in each of the 122 human settlements considered, as reported in the BDINA database ( Table 3 ). ( continued on next page ) Abbreviations: C-City, RT-Rural town, H-Hamlet, V-Village.  Table 1 .      Abbreviations: C-City, RT-Rural town, H-Hamlet, V-Village.

Experimental Design, Materials and Methods
The flooded settlements in the Dosso Region were obtained from the open access national database on flooding-BDINA ( www.inondations-niger.org ) [2] . This source of information provides access to data on flooded sites, loss and damage as collected annually since 1998 by the Early Warning System Coordination Unit through its network of observers on the ground and currently run by the Ministry of Humanitarian Action and Disaster Management. This source has proved to be much richer and more up to date than the global flood databases as Desinventar, the Dartmouth observatory on floods, the EM-DAT, and is therefore used here [3] . Further, 290 flooded settlements were determined between 2011 and 2019 from this primary dataset. These settlements were then identified on satellite images through the geographical coordinates listed in the Human Settlements National Directory, ReNaLoc [4] . For each flooded settlement, the availability of at least three VHR images on GE Pro-were verified over the course of the last 20 years. Therefore, 122 settlements ( Fig. 1 ) were documented by the images captured between 20 01 and 20 09 (20 04 on average), 2010 and 2014 (2012 on average), 2018 and 2020 (2019 on average) ( Fig. 2 ).
Settlements were characterised into urban and rural according to the definition provided by the National Institute of Statistics of Niger. The first category includes cities, i.e. all regional or departmental capitals (41,0 0 0 inhabitants on average). The rural category includes the capitals of rural municipalities (5300 inhabitants on average), villages, the lowest level at which taxes are collected and health and education services are provided (1200 inhabitants on average), and hamlets, simple aggregations of dwellings without administrative functions and community services (500 inhabitants on average) [ 5 , 6 ].
The built-up area of each settlement was visually identified on three dates including all the contiguous developed lots and the road surfaces providing access to these lots. Playgrounds, graveyards, and communication tower lots are also included. Vacant lots, the ones under construction, and the isolated developed lots are excluded when they are separated from the contiguous built-up area more than 60 m away or two standard vacant lots [7] ( Fig. 3 ).
The built-up area is acquired using the "View" and the "Historic imagery" buttons of GE Pro. It was then manually measured using the "Add polygon" GE Pro-tool. A .kmz file is then generated for each built-up area in each year and for the settlement, thereby generating a total of 365 files, freely available at the Mendeley repository. These files are transferred into the QGIS environment to initiate the map production ( Figs. 4 and 5 ).
In rural Niger, adobe buildings prevail. Substituting the earthen roof with a corrugated iron roof is the most common measure to protect houses from heavy rains. Corrugated iron sheets are identified through the visual photointerpretation of VHR satellite images, which are accessible with true colours from GE Pro. The images captured in the months of February-March (twothirds) and September (22%) were not usable. Conversely, the images captured in December-January and June-August have suitable illumination and excellent atmospheric conditions and allowed for the coverage of 40% of the settlements with three images within the period of interest. The colours light grey or light blue indicate corrugated iron roofs. However, as the region is seasonally exposed to dust winds the colour of the roof can be masked. To limit visual photointerpretation errors field inspections were conducted in the villages of Gagila, Sabon Birni, and Takouidawa and in the rural towns of Guéchémé and Tessa ( Fig. 3 ). However, only images captured from unmanned aerial vehicles show without any doubt the roof material [8] .
The effectiveness of roof retrofitting can be appreciated by comparing the widespread use of corrugated iron sheet in building roofs with the number of buildings that have collapsed as a result of pluvial flooding between 2011 and 2019. The latter were taken from the BDINA open access database for the human settlements considered ( Table 3 ). In a few cases flood damage is recorded by the BDINA for clusters of human settlements (three to ten villages and hamlets). To overcame this lack of detail, the number of collapsed buildings for the settlement of interest was estimated by multiplying the number of collapsed buildings in the cluster by the population of the settlement of interest and dividing the result by the total population of the cluster.