Using Sentinel-1 and Google Earth Engine cloud computing for detecting historical flood hazards in tropical urban regions: a case of Dar es Salaam

Figures & data

Figure 1. Study area map. The boundary of Dar es Salaam is delineated from OpenStreetMap. The data of the built up areas is found from GitHub (Dooley et al. 2020).

Figure 2. Climate of the study area. Date source: CLIMATE-DATA.ORG.

Figure 3. Flow diagram of the methodological approaches used.

Figure 4. (A) Footprints of Sentinel-1 (a, b) for May 2021 in Dar es Salaam. (a) Sentinel-1A including three images captured on 05, 17, and 29 May, (b) Sentinel-1B covering about 90% of the city with two images captured on 08 and 20 May, (c) Sentinel-1B covering only southeastern half of the city with three images captured on 03, 15, and 27 May, and (d) overlap of all repeat Sentinel-1A and B images, and (B) temporal distribution of the Sentinel-1A and B images in Dar es Salaam.

Figure 5. Histograms of the single co-polarization (VV) intensity images that cover Dar es Salaam over the period from 01 March–31 May 2021 presented as a sample representation to show the pattern of the sigma VV values (dB) over the images. S1A refers to Sentinel-1 A, and S1B refers to Sentinel-1 B.

Figure 6. Representative threshold charts of the co-polarization (VV) intensity values over the period from 01 March–31 May 2021. The vertical lines indicate the thresholds that separate water and non-water surfaces. The satellite (A/B) is as presented in Figure 4.

Figure 7. Distribution of threshold values of the Sentinel-1 C-band GRD VV polarization images for the period 2016–2021 (n = 55).

Figure 8. Binarized Sentinel-1 images of a few dates that fully cover Dar es Salaam in 2021. The randomly generated validation points are included in the last map.

Table 1. Accuracy assessment result of the binarization of water and land.

	Water	Land	Producer’s accuracy	User’s accuracy
Water	537	21	0.96	0.94
Land	37	469	0.93	0.96

Figure 9. Flooded areas in Dar es Salaam in 2021, presented as representatives.

Table 2. Flood inundated areas (km²) in Dar es Salaam (2016–2021).

	March^a	March	April^a	April	May^a	May	Grand
		Total^b		Total^b		Total^b	Total
2016					3.44	10.32	13.76
2017	35.04	105.11	3.32	9.96	3.14	12.57	169.12
2018	5.50	5.50	4.44	13.31	3.24	6.47	38.47
2019	50.01	200.03	5.79	23.15	3.92	19.58	302.46
2020	2.79	8.38	2.60	10.40	3.19	9.56	36.93
2021	49.32	147.96	2.98	11.94	3.50	17.52	233.23
Total	142.66	466.99	19.13	68.76	20.43	76.02

^aNote: Average flood area per flood event.

^bAverage flood area per event multiplied by the number of events in each month. The number of events is as indicated in Figure 4(B).

Table 3. Historical flood records in Dar es Salaam region reported from various sources (source: FloodList – https://floodlist.com/?s=dar+es+salaam&submit=).

Event	Date	Death	Victiwms/ Displacement	Rainfall depth
1	11–13 April 2014	–	–	–
2	14 April, 2014	10	–	135mm
3	20–25 March 2015	7	5000	91 mm
4	07 May, 2015	12	–	85 mm
5	11–12 May 2015	12	–	52.5 mm
6	20–21 January 2016	–	–	–
7	28 January 2016	–	–	105 mm
8	14–16 April 2018	–	–	81.8, 99.6 mm
9	5–7 May 2019	–	–	144mm
10	06–14 May 2019	2	–	–
11	13 October 2020.	–	–	–
12	04 March, 2020	–	–	–
13	24 April 2021	22	30,005	–

Figure 10. Flood frequency in Dar es Salaam from 2016–2021 rainy seasons.

Figure 11. Flood inundation under different land cover types.

Figure 12. Spatial distribution of flood inundation in relation to different land cover types.

Figure 13. Histogram of VV backscatter values for builtup (a), channelbed (b), floodplain (c), and waterbody (d).

Dooley CA, Leasure DR, Tatem AJ. 2020. Gridded maps of building patterns throughout sub-Saharan Africa, version 1.1. Southampton, UK: University of Southampton.

 Google Scholar

Availability of data and materials

Data are available and can be provided when required.

Code availability

Data were processed using the GEE cloud computing. The GEE codes can be available after accepted for publication.