Quantification of surface water extent and volume in the Inner Niger Delta (IND) over 2000–2022 using multispectral imagery and radar altimetry

Figures & data

Figure 1. Flowchart of the method used.

Figure 2. Inner Niger Delta (IND) study site and location of main cities. In situ station (black circles), ERS-2/Envisat/Saral missions (purple triangles), Sentinel-3A (green crosses) and Sentinel-3B (black crosses) are represented on the map. The Niger and Bani rivers are in blue. Flow direction is indicated by blue arrows. The bottom of the map is the digital elevation model Shuttle Radar Topography Mission (SRTM, freely available on NASA website).

Table 1. Satellite products used in the study.

Mission	Terra	ERS-2 ENVISAT	Saral	Sentinel-3A and 3B	ICESat-2	Global Surface Water Dynamics (Pickens et al. 2020)	Global Surface Water (Pekel et al. 2016)
Instrument	MODIS	Radar Altimeter (RA) Radar Altimeter (RA-2)	AltiKa	Sar Radar Altimeter (SRAL)	Advanced Topographic Laser Altimeter System (ATLAS)	Landsat-5, 7 and 8 missions	Landsat missions
Product	MOD09A1 composite Level-3	GDR Level-2	GDR Level-2	GDR Level-2	ATLAS/ICESat-2 L3A ATL13	GSWD	GSW
Downloading	https://appeears. earthdatacloud. nasa.gov/	https://hydroweb. next.theia-land.fr/	https://hydroweb. next.theia-land.fr/	https://hydroweb. next.theia-land.fr/	https://nsidc.org/ data/icesat-2	https://glad.umd.edu/ dataset/global-surface- water-dynamics	https://global- surface-water.appspot. com/download
Data type	Multispectral imagery	Radar altimetry	Radar altimetry	Radar altimetry	Lidar	Multispectral imagery	Multispectral imagery
Use objective	Flood map	Water level	Water level	Water level	Validation	Validation	Validation
Operation date	2000–now	1995–2003 2002–2012	Since 2013	2016–now 2018–now	2018–now	1999–2021	1984–2021
Space agency	NASA	ESA	CNES, ISRO	ESA	NASA	–	–
Repetitivity (days)	8	35	35	27	91	30	30
Spatial reoslution (m)	500	–	–	–	–	30	30
Spectral indexes	EVI = $a\times \frac{\rho \mathit{\text{NIR}}–\rho \mathit{\text{red}}}{\rho \mathit{\text{NIR}}+b\times \rho \mathit{\text{red}}-c\times \rho \mathit{\text{blue}}+d}$ LSWI = $\frac{\rho \mathit{\text{NIR}}–\rho \mathit{\text{SWIR}}}{\rho \mathit{\text{NIR}}+\rho \mathit{\text{SWIR}}}$	–	–	–	–	NDWI = $\frac{\rho \mathit{\text{NIR}}-\rho \mathit{\text{green}}}{\rho \mathit{\text{NIR}}+\rho \mathit{\text{green}}}$ MNDWI = $\frac{\rho \mathit{\text{green}}-\rho \mathit{\text{SWIR}}}{\rho \mathit{\text{green}}+\rho \mathit{\text{SWIR}}}$	NDVI = $\frac{\rho \mathit{\text{red}}–\rho \mathit{\text{NIR}}}{\rho \mathit{\text{NIR}}+\rho \mathit{\text{SWIR}}}$ Hue-Saturation- Value (HSV)

Figure 3. Time series of surface water extent (km²) calculated using NORMANDIN2018, ZWARTS 2018, PEKEL2016 and PICKENS2020.

Figure 4. Mean flooding duration using the five methods. (a) from NORMANDIN2018 (with mixed pixels, (b) from NORMANDIN2018 (with mixed pixels), (c) from ZWARTS2018, (d) from PEKEL2016, and (e) from PICKENS2020.

Figure 5. Comparison of water level maps from NORMANDIN2018 and ICESat-2, ZWARTS2018 and ICESat-2. For each comparison, (a) N is the number of samples compared averaged used for each comparison, (b) is the determination coefficient R², (c) is the Root mean Square Error (RMSE) in meters and (d) is the bias in meters.

Figure 6. Comparison of water levels measured (m) by ICESat-2, NORMANDIN2018 and ZWARTS2018 along two altimeter tracks, (a) map of the locations of the two altimetry tracks chosen, (b) water levels according to the latitude in the downstream and (c) according to the upstream. For each figure, ICESat-2 points are shown in blue, NORMANDIN2018 in green and ZWARTS2018 in black. For each comparison between ICESat-2 and NORMANDIN2018, N, R² and bias are displayed in green and in black for ICESat-2 and ZWARTS2018 comparison.

Figure 7. Maps of statistical parameters for the water level maps from NORMANDIN2018 (combining surface water extent and radar altimetry) with ICESat-2 data, superimposed on a Google Earth image from ESRI world imagery in QGIS software. (a) N is the number of samples compared, (b) the determination coefficient R², (c) the Root mean Square Error (RMSE, in meters) and (d) the bias (in meters).

Figure 8. Time series of surface water volume (km³) calculated using NORMANDIN2018, ZWARTS2018 PEKEL2016 and PICKENS2020 methods.

Figure 9. Mean seasonal cycle between surface water extent (km²) and surface water volume (km³) using (a) NORMANDIN2018 and (b) ZWARTS2018. The numbers represent the month of the year.

Pickens AH, Hansen MC, Hancher M, Stehman SV, Tyukavina A, Potapov P, Marroquin B, Sherani Z. 2020. Mapping and sampling to characterize global inland water dynamics from 1999 to 2018 with full Landsat time-series. Remote Sens Environ. 243:111792. doi:10.1016/j.rse.2020.111792.

 Web of Science ®Google Scholar

Pekel J-F, Cottam A, Gorelick N, Belward AS. 2016. High-resolution mapping of global surface water and its long-term changes. Nature. 540(7633):418–422. doi:10.1038/nature20584.

 PubMed Web of Science ®Google Scholar